@prefix prodottidellaricerca: . @prefix istituto: . @prefix prodotto: . istituto:CDS072 prodottidellaricerca:prodotto prodotto:ID291482 . @prefix modulo: . modulo:ID8139 prodottidellaricerca:prodotto prodotto:ID291482 . @prefix pubblicazioni: . @prefix unitaDiPersonaleInterno: . unitaDiPersonaleInterno:MATRICOLA15534 pubblicazioni:autoreCNRDi prodotto:ID291482 . @prefix rdf: . @prefix retescientifica: . prodotto:ID291482 rdf:type retescientifica:ProdottoDellaRicerca , prodotto:TIPO1303 . @prefix rdfs: . prodotto:ID291482 rdfs:label "Monitoring post-fire vegetation green vegetation cover dynamics in European Burnt Areas from MODIS scaled NDVI time series (Comunicazione a convegno)"@en . @prefix xsd: . prodotto:ID291482 pubblicazioni:anno "2014-01-01T00:00:00+01:00"^^xsd:gYear . @prefix skos: . prodotto:ID291482 skos:altLabel "
Lorenzo Busetto, Peter Strobl, Tracy Huston Durrant, Roberto Boca, Francesco Boccacci, Andrea Camia, Jesus San Miguel-Ayanz (2014)
Monitoring post-fire vegetation green vegetation cover dynamics in European Burnt Areas from MODIS scaled NDVI time series
in ForestSAT 2014, Riva del Garda, 4-7 November 2014
"^^rdf:HTML ; pubblicazioni:autori "Lorenzo Busetto, Peter Strobl, Tracy Huston Durrant, Roberto Boca, Francesco Boccacci, Andrea Camia, Jesus San Miguel-Ayanz"^^xsd:string ; pubblicazioni:altreInformazioni "http://ocs.agr.unifi.it/index.php/forestsat2014/ForestSAT2014/schedConf/presentations"^^xsd:string ; pubblicazioni:url "http://ocs.agr.unifi.it/index.php/forestsat2014/ForestSAT2014/schedConf/presentations"^^xsd:string ; pubblicazioni:affiliazioni "CNR - IREA (Institute on Electromagnetic Sensing of the Environment); EC JRC - Institute for Environment and Sustainability - Forest Resources and Climate Unit; Italy"^^xsd:string ; pubblicazioni:titolo "Monitoring post-fire vegetation green vegetation cover dynamics in European Burnt Areas from MODIS scaled NDVI time series"^^xsd:string ; prodottidellaricerca:abstract "The objective of this work was the development of a methodology for monitoring green vegetation cover regeneration dynamics on forest burnt areas (BAs) identified at European scale within the EFFIS (European Forest Fire Information System) service, through the analysis of coarse-resolution satellite images.\nThe developed method is based on the use of MODIS NDVI 16-days composite images, at 250 meters spatial resolution (Product MOD13Q1). Mid-summer MODIS images were used to create yearly maps of a novel rescaled version of the NDVI index (NDVI^R) for the 2000-2012 period. NDVI^R is expressed as the percentage difference between the original NDVI of a pixel, and the median NDVI of neighboring non-burnt pixels belonging to the same Land Cover (LC) class. Its use allows to reduce the problems in the analysis of post-fire dynamics related to interannual variations unrelated to fire effects. NDVI^R time series of more than 1500 large burnt areas mapped in Europe in the 2003-2011 period were extracted and used to analyze the speed of recovery of green vegetation cover in Forest and Other Wooded Land LC classes. The recovery time (RT) was estimated using a Wilcoxon Signed Rank Test to determine the number of years required for NDVI^R to return to pre-fire levels. Statistical analyses based on Survival Analysis techniques were then conducted to i) characterize the typical distribution of estimated RT for the different LC classes on the basis of a Kaplan-Meyer analysis, and ii) analyze the relationships between RT and different predictors related to characteristics and climate of the different burnt areas (e.g., Burn Severity, LC Class, Burnt Area size, Fire Year, Global Aridity Index), through the use of an Accelerated Failure Time (AFT) regression model.\nResults showed that the developed methodology makes it possible to effectively visualize, analyze and compare the post-fire recovery dynamics of different BAs, and to highlight differences in the recovery speeds of different LC classes affected by fire within the same burnt area. The software implementation of the methodology allows the user to update the analysis easily at the end of each fire season exploiting the updated Burnt Areas' database and MODIS time series. The statistical analysis on 2003-2011 BAs showed that the distribution of the estimated RT varies with LC class, with Broadleaved forests showing the fastest recovery and Coniferous forests the slowest. Results of the AFT regression analysis highlighted a good relationship between RT and the predictors (Nagelkelke Pseudo R^2 = 0.75). RT was found to be mostly influenced by Burn Severity, as estimated on the basis of the median reduction of NDVIR in the year following the fire. Besides this main control, statistically significant effects were also identified for the Global Aridity Index and the Burnt Area size, leading to slower recovery in larger and more arid BAs. The analysis of regression coefficients obtained for the LC Class predictor suggested that, when other parameters are kept constant, recovery time can be expected to be slightly faster in Broadleaved Forest and Schlerophyllus Vegetation areas, while Coniferous Forests show the longer recovery times. In the future, as the length of the time series and the number of analyzed burnt areas increases, further insights may be gained into the post-fire regeneration dynamics of green vegetation cover in Europe, and on their relationships with ecological and climatic factors." ; prodottidellaricerca:prodottoDi istituto:CDS072 , modulo:ID8139 ; pubblicazioni:autoreCNR unitaDiPersonaleInterno:MATRICOLA15534 .